Browser AI Agent

AI-powered browser automation with Ollama, Gemini, OpenCode Go, or ChatGPT / Codex

Your browser, run by AI > Browser AI Agent is an AI assistant that lives in Chrome's side panel. Give it a task in plain English and watch it work. It plans the steps, clicks and types on the page, reads content, and checks the result before telling you it is done. Examples: - "Summarize this article" - "Search for the latest AI news and list the top results" - "Fill out this form with the details below" - "Find the price of the first product and compare it with the second" Choose your AI - Ollama: free and fully local. Nothing leaves your machine. - Google Gemini: bring your own API key. - OpenCode Go: one subscription for a range of current models. - ChatGPT / Codex: sign in with your ChatGPT account. The extension loads the model list from your selected provider, so you pick from what your account can actually use. ### Built to be careful - Sensitive actions such as payments, logins, deletes, submits, and page scripts ask for your approval first. - The agent verifies outcomes instead of assuming clicks worked. When it cannot verify, it says so instead of claiming success. - An action with an unknown result (for example a click or script that did not report back) pauses the task. You inspect the page, then choose whether to retry. - Every task has action and time limits, so it cannot spin forever. ### Built to recover - Remembers verified selectors per site, so repeat tasks on the same site get faster. - Tasks are checkpointed. Close the panel, switch tabs, or restart the browser and a paused task can resume from a fresh look at the page. - Pause, resume, and cancel any running task from the side panel. ### Visual when needed - Models with image support can request a viewport screenshot to interpret icons, canvas content, or unclear page layouts.